Dynamic Regret of Adversarial Linear Mixture MDPs

Authors: Long-Fei Li, Peng Zhao, Zhi-Hua Zhou

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We study reinforcement learning in episodic inhomogeneous MDPs with adversarial full-information rewards and the unknown transition kernel. We consider the linear mixture MDPs whose transition kernel is a linear mixture model and choose the dynamic regret as the performance measure. Denote by d the dimension of the feature mapping, H the length of each episode, K the number of episodes, PT the non-stationary measure, we propose a novel algorithm that enjoys an e O H4(K + PT )(1 + PT ) dynamic regret under the condition that PT is known, which improves previously best-known dynamic regret for adversarial linear mixture MDP and adversarial tabular MDPs. We also establish an Ω HK(H + PT ) lower bound, indicating our algorithm is optimal in K and PT .
Researcher Affiliation Academia Long-Fei Li, Peng Zhao, Zhi-Hua Zhou National Key Laboratory for Novel Software Technology, Nanjing University, China School of Artificial Intelligence, Nanjing University, China {lilf, zhaop, zhouzh}@lamda.nju.edu.cn
Pseudocode Yes Algorithm 1 POWERS-Fix Share; Algorithm 2 POWERS-Fix Share-On E
Open Source Code No The paper does not contain any statement about releasing source code or provide a link to a code repository for the methodology described.
Open Datasets No The paper is theoretical and does not conduct experiments on a specific dataset. Thus, there is no mention of a public dataset or its accessibility.
Dataset Splits No The paper is theoretical and does not conduct experiments with datasets, therefore it does not provide any training/validation/test splits.
Hardware Specification No The paper describes theoretical work and algorithm design; it does not report on empirical experiments, therefore no hardware specifications are mentioned.
Software Dependencies No The paper describes theoretical work and algorithm design; it does not report on empirical experiments requiring specific software dependencies with version numbers.
Experiment Setup No The paper is theoretical and focuses on algorithm design and analysis, not empirical experimentation. The section titled 'Problem Setup' describes the mathematical model, not a practical experimental configuration. No hyperparameters or system-level training settings are provided.